1. Field of the Invention
The invention relates to a method and apparatus for smoothing an image signal, more particularly to an image smoothing apparatus which employs a mean value image smoothing technique to smoothen a two-dimensional image signal.
2. Description of the Related Art
Various image signal smoothing techniques which operate on spatial or frequency domain have been developed in order to decrease or eliminate the presence of noise. One example of a conventional image signal smoothing technique is the mean value image smoothing. In the mean value image smoothing, every pixel of an image signal is replaced by a mean value of the image signal at a plurality of pixels in the neighborhood or vicinity of said pixel. The mathematical equation for finding a smoothed value g(x,y) of a pixel f(x,y) of an image signal which has a total number of N.times.N pixels is as follows: ##EQU1##
wherein,
x=0, 1, 2, 3, . . . N-1; PA1 y=0, 1, 2, 3, . . . N-1;
"S" is a predetermined region (X,Y) which contains a plurality of pixels in the vicinity of the pixel f(n,m); and PA2 M=the number of pixels in "S".
Note that mean value image smoothing can be accomplished in a horizontal direction, in a transverse direction, or in a combination of both directions. Three equations which represent three types of mean value image smoothing techniques may be derived from equation (1) and are given as follows:
1. Horizontal mean value image smoothing: EQU g(x,y)=1/2[f(n,m)+f(n+1,m)] (2)
2. Transverse mean value image smoothing: EQU g(x,y)=1/2[f(n,m)+f(n,m+1)] (3)
3. Center point mean value image smoothing: EQU g(x,y)=1/4[f(n,m)+f(n+1,m)+f(n,m+1)+f(n+1,m+1)] (4)
Conventional image smoothing apparatus which employ the mean value smoothing technique are relatively slow and are not ideal for processing a large amount of image data because they do not employ pipeline processing. Furthermore, the conventional image smoothing apparatus require the use of large blocks of memory space when in operation.